Hello everyone,

has anyone else encountered the problem that when running measurement 4 on a custom dataset the error message AttributeError: 'ResNet' object has no attribute 'classifier' is generated(file gender_based.py)? When printing the module there is also no attribute classifier. How can I resolve this error, am I using the wrong .pth.tar maybe?

This is the output of model:

ResNet(
(conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
(layer1): Sequential(
(0): BasicBlock(
(conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(1): BasicBlock(
(conv1): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(layer2): Sequential(
(0): BasicBlock(
(conv1): Conv2d(64, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(downsample): Sequential(
(0): Conv2d(64, 128, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): BasicBlock(
(conv1): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(layer3): Sequential(
(0): BasicBlock(
(conv1): Conv2d(128, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(downsample): Sequential(
(0): Conv2d(128, 256, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): BasicBlock(
(conv1): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(layer4): Sequential(
(0): BasicBlock(
(conv1): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(downsample): Sequential(
(0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
(1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(1): BasicBlock(
(conv1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
(bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
)
(avgpool): AdaptiveAvgPool2d(output_size=(1, 1))
(fc): Linear(in_features=512, out_features=365, bias=True)
)

Thank you very much and best regards

Hi Angelina,

just a quick question. How should the hierarchy visualization be interpreted?

Thank you very much in advance! :)

All the best,
Nastassia

I want to use it to check the bias of my training data（face dataset）. Do I need to preprocess the dataset with some annotators in advance ?

Hello,

I think the download bash script should be updated.
Using the current script caused the error: _pickle.UnpicklingError: invalid load key, '<'.

By manually downloading from this link rather than the one from download.sh I resolved the error.

Just as a tip for future users. :)

Thank you very much for providing this tool.

Best regards,
Nastassia

Hi,

I have a short question regarding the bounding boxes. In measurement 3 (M3) you use bounding boxes from gender and from anns. However, I don't quite understand the difference between those.
In my understanding, gender is passed the person bounding box. But which bounding box is passed to anns?

Thank you very much in advance.

All the best,
Nastassia

I am getting following error while loading coco dataset,
FileNotFoundError: [Errno 2] No such file or directory: 'Data/Coco/2014data/annotations/instances_train2014.json'
can anyone please provide the link to download above files?

Thanks in advance.

Hello,
I have a short question about using your custom dataset.
I have followed the instructions about adapting TemplateDataset to your specific dataset.
The dataset I am working on contains emotion labels for each respective person in the image - multiple people can be portrayed in one image, thus multiple emotion labels per image exist.
How should this case be treated?
So far, I tried this:

      
 def from_path(self, file_path):

             ......

        annotation_mat_file_name = annotation_path +image_id.replace('.jpg', '.mat')
        mat_file = loadmat(annotation_mat_file_name) 	

        people = mat_file['people']    


        gender_info = []
        image_anns = []
        for p in range(0,len(people)-1):
            person_bbox = bbox = people['person'][p]['bbx']
            gender = people['person'][p]['gender']
            new_ann = {'bbox': person_bbox, 'label': people['person'][p]['cats']}
            image_anns.append(new_ann)
            gender_info.append([gender, self.biggest_bbox])
        country = None

        scene_group = self.scene_mapping[(self.img_folder + file_path)] # optional
        anns = [image_anns, gender_info, [country], (self.img_folder + file_path), scene_group]

        return image, anns

I saved each labelling per image in a .mat file. People is a struct with numofpersons entries, each containing a bounding box of the person, its gender, labels and other entries.
The problem is the following: The lists inside of 1.pkl (and my guess is, also for the other measurements) are empty. Presumably, because the annotations I pass during from_path are incorrect.

Do you have any suggestions on how to proceed? Should I save one .mat file per annotated person and thus treat each person and its corresponding bounding box as a separate image?

I will be very greatful for your input on this matter and I am very looking forward to your reply.

Best regards,
Nastassia

Hi,
it's me again.

Just as a quick tip for other users/you. :)
In datasets.py you stated in the TemplateDataset in the method from_path:

gender = None # optional, we have used 0 for female and 1 for male when these labels exist

So, I followed the instruction and have set:
if gender == "Female": binary_gender = 0 if gender == "Male": binary_gender = 1

I calculated the number of Male and Female persons depicted in the images beforehand, so I know how many of them there are. Unfortunately the number for Male and Female persons is exactly switched in the first measurement (M0) which leads me into believing that you used

0:Male

1:Female 

labeling instead.

No critique, just a friendly tip for you/your future users. Since running the measurement 1 took a bit more than two hours for me and now I will rerun it, I want to prevent this for your users. :)

All the best to you and best regards,
Nastassia

Hi,

First of all thanks a lot for releasing wonderful tool. Kudos.
I am struggling with running the tool properly on COCO dataset as this seems to be a missing link-

gender_data = pickle.load(open('Data/Coco/2014data/bias_splits/train.data', 'rb'))

Also, I am not able to understand what is happening in the next line:

self.gender_info = {int(chunk['img'][15:27]): chunk['annotation'][0] for chunk in gender_data}

Can you please guide me with the path to download this pickle file which is acceptable by the code.

Hi Angelina,

sorry, it's me again!
I have a short question regarding M1. You compute the distance from the person to the image center as:
distance = np.linalg.norm(person_center - img_center)
However, you set:
img_center = np.array([.5, .5])

Maybe I am wrong, but did you maybe forget to multiply with the image size itself? Since the person_center is given in pixel values.

Please correct me if I am wrong, I was just wondering. :)

Thanks so much,
Nastassia

Hi,

I have a short question regarding the bounding boxes. In measurement 3 (M3) you use bounding boxes from gender and from anns. However, I don't quite understand the difference between those.
In my understanding, gender is passed the person bounding box. But which bounding box is passed to anns?

Thank you very much in advance.

All the best,
Nastassia

Hi,

when trying to run the Object Analysis notebook on the Coco dataset I am getting the following error in the "Analyses" section of "(M7) Metric: Size and Distance from Center of Supercategories":

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-38-6ffb36b669b2> in <module>
      1 if first_pass:
----> 2     object_size(cat_to_ent[0][1], None)
      3 ui = HBox(all_things)
      4 out = widgets.interactive_output(object_size, {'object_class': object_class_widget, 'sizes': sizes_widget})
      5 display(ui, out)

<ipython-input-37-46aecec2feb8> in object_size(object_class, sizes)
    144     these_instances = np.concatenate(instances_per, axis=0)
    145     scenes_per = np.array([dataset.from_path(filepath)[1][4] for filepath in filepaths])
--> 146     these_scenes = np.concatenate(scenes_per, axis=0)
    147     num, counts = np.unique(these_instances, return_counts=True)
    148     num = np.array([categories.index(nu) for nu in num])

<__array_function__ internals> in concatenate(*args, **kwargs)

ValueError: zero-dimensional arrays cannot be concatenated

<Figure size 432x288 with 0 Axes>

The issue is, that the variable scenes_per is simply a list of None's:

[None None None None None None None None None None None None None None
 None None None None None None None None None None None None None None
 None None None None None None None None None None None]

printing dataset.from_path(filepath)[1] for one filepath in filepaths results in:

[[{'bbox': [0.23233333333333334, 0.999, 0.22865625, 1.0], 'label': 1}, {'bbox': [0.28827083333333337, 0.6980625, 0.6325000000000001, 0.88925], 'label': 88}, {'bbox': [0.003, 0.23970833333333333, 0.9039375, 0.987078125], 'label': 65}], [1, [0.23233333333333334, 0.999, 0.22865625, 1.0]], [0], '/n/fs/visualai-scr/Data/Coco/2014data/train2014/COCO_train2014_000000176176.jpg', None]

I verified, that the image at the given path exists. Can anyone help me with this issue? Thanks in advance!

Best regards,
Tobias

Traceback (most recent call last):
File "D:\Bias\Image\Bias_in_image\revise-tool\main_measure.py", line 90, in
main()
File "D:\Bias\Image\Bias_in_image\revise-tool\main_measure.py", line 86, in main
index_to_measurement[meas](dataloader, args)
File "D:\Bias\Image\Bias_in_image\revise-tool\measurements\object_based.py", line 176, in obj_scn
group_mapping = dataloader.dataset.group_mapping
File "C:\Users\nipon.chanda.conda\envs\bias_env\lib\site-packages\torch\utils\data\dataset.py", line 83, in getattr
raise AttributeError
AttributeError